HuaYuexia
Results
3
issues of
HuaYuexia
Can you talk about how the method comparison in Section 4.2 inserted these methods into Faster R-CNN?
Loss is
4
请问qwen2.5vl的编码方式是只能将图片token编码到提问text的起始位置嘛? 如果有多张图片,在add_vision_id=True已经设置的情况下。请问应该怎么把图像id添加到prompt中去呢,比如我有两张图片,怎么表示呢? '图像1 are {} images extracted evenly from a video. ' + '图像2 are {} images extracted evenly from another video. ' 如何把这个图像1,图像2用图像id表示呢?才能让图像token插入到我想要的位置上